Image Retrival Based On Edge Structure Characteristic

This text put forward a kind of method that inspects the image with the edge structure characteristic.Overpass calculate matrix of element space distributing. That is in the part-edge image we can get the eigenvetor. The calculate way that this text put forward can withdraw to imbed in the information in the edge effectively. It is proved in the experiment that such structure characterization is good for catching edge imformation, in order to elevated image retrival capability.This text passes to carry on several demarcations to the edge diagram, gathering the image to a set that means the shape orders the space relation of constant quantity characteristic. Its characteristics is that the edge withdraw not means of be the target partition, but see only as reflect several low level characteristics that the structure distribute of image. That method avoided because of the difficulty of the target of image comprehension but request partition,reflecting the space structure of the image to distribute availably at the same time. The picture divide the line the choice of the dimensions to has been a hard nut to crack, this text puts forward to choose the excellent dimensions method for turn under the partial biggest standard of entropy, electing to choose from the edge image divide the line behind a little bit some, the relation conduct and actions that withdraws the of the shape point is constantly the immutability eigenvetor. This method reduced an information redundancy of ash availably, For including the general to have no stipulation a view or have complications the image of the structure identified to provide a kind of dependable according to structure of the image contents description method.It is proved in the experiment that choose thus of it is meaningful that image divide the line to withdraw with characteristic, can raise the inspectional usefulness.